Have an idea?

Visit Sawtooth Software Feedback to share your ideas on how we can improve our products.

Correlate part-worth utilities with other variable of interest

I was wondering whether it would make sense to correlate part-worth utilities and/or  importance scores to some other variable of interest (e.g age). For example: if I calculate the correlation between part-worth values of the color of a product and age I may find a positive correlation for age and blue and a negative for age and red. But do part worth utilities (the data type) allow for such a calculated to be meaningfully interpreted?

Many thanks in advance!
asked Nov 13, 2019 by Eve

1 Answer

0 votes
Yes, you can treat utilities and importances as continuous variables.  Importances are already normalized per respondent to sum to 100.  But, raw HB utilities are potentially on quite different scales for different people depending on the consistency with which they answer the conjoint questions.  So, if using HB utilities in something like cluster or correlation analysis, you should use the "zero-centered Diffs" version of those utilities, which are normalized per person.
answered Nov 13, 2019 by Bryan Orme Platinum Sawtooth Software, Inc. (174,440 points)